Delay Cost And The Decision Price

A portfolio manager decides at 9:31am to buy a stock trading at $50.00. The order doesn't reach a trader's desk, get sized, and start executing until 9:38am, by which point the stock has drifted to $50.15 — with no market impact from this order involved at all, simply the ordinary movement of the market during those seven minutes. That $0.15 is a real cost of the trade, even though no order was even in the market yet when it accrued. This is delay cost: the price movement between when a decision is made and when execution actually begins, and it's one of the clearest illustrations of why "trading cost" has to be measured against the moment of decision, not the moment of the first fill.

Why the decision price is the right benchmark

If you only measured execution quality against the price when the order started trading, delay cost would be invisible by construction — you'd be comparing execution to a price that had already absorbed the drift. Perold's implementation shortfall framework fixes this by anchoring everything to the decision price: the price prevailing at the moment the investment decision was made. Every dollar of difference between that anchor and the final realized outcome — whether from delay before trading starts, market impact while trading, or the opportunity cost of any portion never filled at all — counts as a cost of implementing the decision, because from the portfolio's perspective, that's exactly what it is: the gap between the paper decision and the real, executed result.

Delay cost specifically isolates the piece that accrues before a single share trades. It can come from operational lag (an order sitting in a queue, waiting for trader attention, moving between systems) or from a deliberate choice to wait for better conditions — either way, the market doesn't pause for the trader's internal process, and the resulting drift is a cost regardless of its cause.

Worked example: splitting a shortfall into delay and the rest

A manager decides to buy 50,000 shares at a decision price of $80.00. Execution doesn't begin for 20 minutes, by which time the price has moved to $80.10 — this is the delay. Trading then proceeds and the order fully fills at an average price of $80.28, reflecting both the $80.10 starting point and additional market impact and further drift while working the order. Total implementation shortfall per share:

$$80.28 - 80.00 = 0.28 ,$$

i.e. $0.28 per share. Splitting it: delay cost is $80.10 - 80.00 = 0.10$, i.e. $0.10 per share, and the remaining $80.28 - 80.10 = 0.18$ is attributable to impact and drift during actual execution, i.e. $0.18 per share. On 50,000 shares, that's $50{,}000 \times 0.10 = 5{,}000$, i.e. $5,000 lost purely to delay before the order even started, versus $50{,}000 \times 0.18 = 9{,}000$, i.e. $9,000 from the execution itself — delay alone accounts for over a third of the total shortfall here, purely from the seven-plus-minute gap between decision and action.

What this means in practice

Delay cost puts accountability on the parts of the process before a trader even sends an order — how quickly decisions reach the desk, how orders are prioritized and queued — which a naive cost analysis anchored only to the start-of-trading price would completely miss. Funds serious about minimizing total trading cost measure and manage delay explicitly, because it's often a purely operational cost with no offsetting benefit, unlike market impact, which at least buys immediacy.